Transferring a brain mask from T1-weighted space into native FLAIR space is a common preprocessing step, but T1-to-FLAIR rigid registration can be unstable. The authors propose A3PreCore, a six-degree-of-freedom registration that aligns the full bias-corrected T1 image to native FLAIR using normalized mutual information while preserving the existing T1-space mask. They also introduce E-QC, a multidimensional quality-control framework that assesses mask extent, FLAIR intensity support, boundary support, inferior morphology, and pairwise geometric disagreement.
In 78 participants with paired T1 and FLAIR scans, A3PreCore produced no gross rigid-body misregistrations, compared with 21/78 (26.9%) for the standard fsl_anat pipeline (exact two-sided McNemar p = 9.54 × 10⁻⁷). In the 57 participants where fsl_anat did not grossly fail, A3PreCore remained closely concordant (median Dice = 0.993; median paired ΔNMI = −0.00007, p = 0.429). Against BET, A3PreCore was broadly concordant (median Dice = 0.955) and showed less inferior and extracranial extension in representative comparisons.
Method-blinded neuroscientists ranked A3PreCore first in 35 of 40 participants (87.5%). The authors also report that E-QC exposed severe orientation-related disagreements that global descriptors and centroid displacement alone did not reveal, suggesting it could serve as a complementary tool for participant-specific geometric quality control.